Merge pull request #134 from carlosp420/vouchers_per_gene

Vouchers per gene

voseq/blast_ncbi/utils.py

@@ -28,8 +28,11 @@ def do_blast(self):
         """
         Does a blast against NCBI and saves returned XML file to local disk.
         """
-        fasta_string = open(self.query_file).read()
+        with open(self.query_file) as handle:
+            fasta_string = handle.read()
+
         result_handle = NCBIWWW.qblast('blastn', 'nt', fasta_string)
+
         with open(self.output_file, 'w') as writer:
             writer.write(result_handle.read())
         result_handle.close()

voseq/create_dataset/dataset.py

@@ -107,12 +107,12 @@ def get_number_of_genes_for_taxa(self, partitions):
         gene_code = ''
         for item in partitions[0]:
             if item.startswith('\n'):
-                    if self.file_format == 'NEXUS' or self.file_format == 'PHY':
-                        gene_code = item.strip().replace('[', '').replace(']', '')
-                        continue
-                    if self.file_format == 'TNT':
-                        gene_code = 'dummy'
-                        continue
+                if self.file_format == 'NEXUS' or self.file_format == 'PHY':
+                    gene_code = item.strip().replace('[', '').replace(']', '')
+                    continue
+                if self.file_format == 'TNT':
+                    gene_code = 'dummy'
+                    continue
             if gene_code != '':
                 entry = re.sub('\s+', ' ', item)
                 voucher, sequence = entry.split(' ')

voseq/stats/management/commands/create_stats.py

@@ -3,13 +3,17 @@
 from public_interface.models import Vouchers
 from public_interface.models import Sequences
 from stats.models import Stats
+from stats.models import VouchersPerGene
 
 
 class Command(BaseCommand):
     help = 'Extracts total number of orders, families, genera, etc. from ' \
-           'our database.'
+           'our database. Also counts the number of vouchers for each of ' \
+           'our genes.'
 
     def handle(self, *args, **options):
+        self.count_vouchers_per_gene()
+
         queryset = Vouchers.objects.all()
 
         orders = set()
@@ -52,3 +56,18 @@ def handle(self, *args, **options):
                 'sequences': num_sequences,
             }
         )
+
+    def count_vouchers_per_gene(self):
+        genes = Sequences.objects.all().values('gene_code').distinct()
+
+        model_objects = []
+
+        pk_index = 1
+        for gene in genes:
+            voucher_count = Sequences.objects.filter(gene_code=gene['gene_code']).count()
+            gene['voucher_count'] = voucher_count
+            model_objects.append(VouchersPerGene(id=pk_index, **gene))
+            pk_index += 1
+
+        VouchersPerGene.objects.all().delete()
+        VouchersPerGene.objects.bulk_create(model_objects)

voseq/stats/models.py

@@ -2,8 +2,7 @@
 
 
 class Stats(models.Model):
-    """
-    Statistics about number of vouchers, photos, etc to show in page footer.
+    """Statistics about number of vouchers, photos, etc to show in page footer.
     """
     vouchers = models.IntegerField(
         help_text='Number of records, or vouchers.'
@@ -15,3 +14,10 @@ class Stats(models.Model):
     genera = models.IntegerField()
     species = models.IntegerField()
     sequences = models.IntegerField()
+
+
+class VouchersPerGene(models.Model):
+    """Number of vouchers that have sequences for each of our genes.
+    """
+    gene_code = models.CharField(max_length=100)
+    voucher_count = models.IntegerField()

voseq/stats/tests.py

@@ -2,6 +2,7 @@
 from django.core.management import call_command
 
 from stats.models import Stats
+from stats.models import VouchersPerGene
 
 
 class TestCustomCommand(TestCase):
@@ -16,3 +17,13 @@ def test_create_stats(self):
         res = Stats.objects.get(id=1)
         expected = 10
         self.assertEqual(expected, res.vouchers)
+
+    def test_count_vouchers_per_gene(self):
+        call_command('create_stats')
+        res = VouchersPerGene.objects.all().values('gene_code', 'voucher_count')
+        for i in res:
+            if i['gene_code'] == 'COI':
+                self.assertTrue(i['voucher_count'] == 2)
+
+            if i['gene_code'] == '16S':
+                self.assertTrue(i['voucher_count'] == 1)

voseq/view_genes/templates/view_genes/index.html

@@ -18,6 +18,15 @@ <h3>Existing genes:</h3>
                 <b><a href="/genes/{{ i.gene_code }}/">{{ i.gene_code }}</a></b>
                 <i>{{ i.description }}</i>
                 &mdash; {{ i.length|intcomma }} bp
+
+                <span class="badge">
+                {% if i.voucher_count == 1 %}
+                  <b>{{ i.voucher_count }} voucher</b>
+                {% else %}
+                  <b>{{ i.voucher_count }} vouchers</b>
+                {% endif %}
+                </span>
+
               </li>
               {% endfor %}
 

voseq/view_genes/tests/test_views.py

@@ -10,12 +10,20 @@ def setUp(self):
         cmd = 'migrate_db'
         call_command(cmd, *args, **opts)
 
+        call_command('create_stats')
+
         self.client = Client()
 
     def test_genes(self):
         response = self.client.get('/genes/')
         self.assertEqual(200, response.status_code)
 
+    def test_genes_with_number_of_vouchers(self):
+        response = self.client.get('/genes/')
+        expected = '2 vouchers'
+        result = str(response.content)
+        self.assertTrue(expected in result)
+
     def test_view_gene(self):
         response = self.client.get('/genes/wingless/')
         self.assertEqual(200, response.status_code)

voseq/view_genes/views.py

@@ -2,21 +2,39 @@
 
 from core.utils import get_version_stats
 from public_interface.models import Genes
+from stats.models import VouchersPerGene
 
 
 def index(request):
     version, stats = get_version_stats()
 
-    queryset = Genes.objects.all()
+    voucher_count = get_voucher_count()
+
+    queryset = Genes.objects.all().values()
+    result = []
+    for i in queryset:
+        gene_code = i['gene_code']
+        i['voucher_count'] = voucher_count[gene_code]
+        result.append(i)
+
     return render(request, 'view_genes/index.html',
                   {
-                      'result': queryset,
+                      'result': result,
                       'version': version,
                       'stats': stats,
                   },
                   )
 
 
+def get_voucher_count():
+    voucher_count = {}
+    queryset = VouchersPerGene.objects.all().values('gene_code', 'voucher_count')
+    for i in queryset:
+        gene_code = i['gene_code']
+        voucher_count[gene_code] = i['voucher_count']
+    return voucher_count
+
+
 def gene(request, gene_code):
     version, stats = get_version_stats()
 

voseq/voseq/settings/base.py

@@ -16,9 +16,6 @@
 # Quick-start development settings - unsuitable for production
 # See https://docs.djangoproject.com/en/1.7/howto/deployment/checklist/
 
-# SECURITY WARNING: keep the secret key used in production secret!
-SECRET_KEY = 'hola'
-
 
 ALLOWED_HOSTS = []
 

voseq/voseq/settings/testing.py

@@ -4,6 +4,9 @@
 
 
 print('Testing')
+
+SECRET_KEY = 'hola'
+
 TESTING = len(sys.argv) > 1 and sys.argv[1] == 'test'
 
 DATABASES = {

voseq/voseq/settings/travis.py

@@ -3,18 +3,8 @@
 fails most of the time due to network problems.
 """
 
-import sys
+from .testing import *
 
-from .base import *
 
-
-print('Testing in Travis')
-TESTING = len(sys.argv) > 1 and sys.argv[1] == 'test'
+print('in Travis')
 TRAVIS = True
-
-DATABASES = {
-    'default': {
-        'ENGINE': 'django.db.backends.sqlite3',
-        'NAME': 'test.db',
-    }
-}